metagenomic evidence for reciprocal particle exchange ... · smith et al. columbia river estuarine...

ORIGINAL RESEARCHpublished: 01 October 2015

doi: 10.3389/fmicb.2015.01074

Edited by:Hongyue Dang,

Xiamen University, China

Reviewed by:Jennifer F. Biddle,

University of Delaware, USAIan Salter,

Alfred Wegener Institute, Germany

*Correspondence:Holly M. Simon,

Center for Coastal MarginObservation and Prediction

and Institute of Environmental Health,Oregon Health & Science University,3181 SW Sam Jackson Park Road,

Portland, OR 97239, [email protected]

Specialty section:This article was submitted to

Aquatic Microbiology,a section of the journal

Frontiers in Microbiology

Received: 27 May 2015Accepted: 18 September 2015

Published: 01 October 2015

Citation:Smith MW, Davis RE, Youngblut ND,

Kärnä T, Herfort L, Whitaker RJ,Metcalf WW, Tebo BM, Baptista AM

and Simon HM (2015) Metagenomicevidence for reciprocal particle

exchange between the mainstemestuary and lateral bay sediments

of the lower Columbia River.Front. Microbiol. 6:1074.

doi: 10.3389/fmicb.2015.01074

Metagenomic evidence for reciprocalparticle exchange between themainstem estuary and lateral baysediments of the lower ColumbiaRiverMaria W. Smith1, Richard E. Davis1, Nicholas D. Youngblut2, Tuomas Kärnä1,Lydie Herfort1, Rachel J. Whitaker3, William W. Metcalf3, Bradley M. Tebo1,António M. Baptista1 and Holly M. Simon1*

1 Center for Coastal Margin Observation and Prediction and Institute of Environmental Health, Oregon Health & ScienceUniversity, Portland, OR, USA, 2 Department of Crop and Soil Sciences, Cornell University, Ithaca, NY, USA, 3 Department ofMicrobiology, University of Illinois at Urbana-Champaign, Urbana, IL, USA

Lateral bays of the lower Columbia River estuary are areas of enhanced waterretention that influence net ecosystem metabolism through activities of their diversemicrobial communities. Metagenomic characterization of sediment microbiota fromthree disparate sites in two brackish lateral bays (Baker and Youngs) produced∼100 Gbp of DNA sequence data analyzed subsequently for predicted SSU rRNAand peptide-coding genes. The metagenomes were dominated by Bacteria. A largecomponent of Eukaryota was present in Youngs Bay samples, i.e., the inner baysediment was enriched with the invasive New Zealand mudsnail, Potamopyrgusantipodarum, known for high ammonia production. The metagenome was also highlyenriched with an archaeal ammonia oxidizer closely related to Nitrosoarchaeum limnia.Combined analysis of sequences and continuous, high-resolution time series ofbiogeochemical data from fixed and mobile platforms revealed the importance of large-scale reciprocal particle exchanges between the mainstem estuarine water columnand lateral bay sediments. Deposition of marine diatom particles in sediments nearYoungs Bay mouth was associated with a dramatic enrichment of Bacteroidetes (58%of total Bacteria) and corresponding genes involved in phytoplankton polysaccharidedegradation. The Baker Bay sediment metagenome contained abundant Archaea,including diverse methanogens, as well as functional genes for methylotrophy andtaxonomic markers for syntrophic bacteria, suggesting that active methane cyclingoccurs at this location. Our previous work showed enrichments of similar anaerobictaxa in particulate matter of the mainstem estuarine water column. In total, our resultsidentify the lateral bays as both sources and sinks of biogenic particles significantlyimpacting microbial community composition and biogeochemical activities in theestuary.

Keywords: metagenome analysis, Columbia River estuary, lateral bay sediments, particle exchange, methanecycling

Frontiers in Microbiology | www.frontiersin.org 1 October 2015 | Volume 6 | Article 1074

http://www.frontiersin.org/Microbiology/

http://www.frontiersin.org/Microbiology/editorialboard

http://www.frontiersin.org/Microbiology/editorialboard

http://dx.doi.org/10.3389/fmicb.2015.01074

http://creativecommons.org/licenses/by/4.0/

http://dx.doi.org/10.3389/fmicb.2015.01074

http://crossmark.crossref.org/dialog/?doi=10.3389/fmicb.2015.01074&domain=pdf&date_stamp=2015-10-01

http://journal.frontiersin.org/article/10.3389/fmicb.2015.01074/abstract

http://loop.frontiersin.org/people/75948/overview











http://www.frontiersin.org/

http://www.frontiersin.org/Microbiology/archive

Smith et al. Columbia River estuarine sediment metagenomes

Introduction

Coastal zones worldwide receive significant nutrient inputs fromland through rivers. These inputs are modulated by estuaries(Jickells, 1998; Satinsky et al., 2014) with highly variable physicaldynamics that are controlled by tides and by river and oceanforcing. Tides and external forcing also influence the compositionof estuarine biota and the nature of nutrient fluxes (Blum andMills, 2012). Organic matter inputs to estuaries are providedby terrigenous sources (runoff from watersheds transportedby rivers), submerged aquatic vegetation of fresh/salt watermarshes, and planktonic organisms developing in the river-to-ocean continuum (Crump et al., 2012). These inputs consist oflabile dissolved and more recalcitrant particulate organic matter(DOM and POM, respectively; Simon et al., 2002; Jiao et al.,2010).

Estuarine water column particles include mineral, biogenicand anthropogenic components from both aquatic and terrestrialenvironments. These particles are the foci of biologic activitythrough colonization by bacteria, protozoa, and metazoainvolved in nutrient recycling and organic matter decomposition(Zimmermann-Timm, 2002; Crump et al., 2004, 2012; Grossartet al., 2007; Satinsky et al., 2014). Tidally driven cycles of particledeposition and resuspension provide for continuous chemicalexchanges between the aqueous phase, suspended phase, andriverbed (Turner and Millward, 2002). Multiple studies haveindicated that particle-associated microorganisms are distinct incharacteristics related to growth and activities from those in free-living fractions (DeLong et al., 1993; Crump et al., 1999; Simonet al., 2002, 2014; Smith et al., 2013; Satinsky et al., 2014).

The Columbia River estuary is a particle-rich environment(Haertel et al., 1969) that receives a freshwater supply of over 10million tons of sediment per year to the estuary (Sherwood et al.,1984). In the lower Columbia River estuary, tidally modulated,hydrodynamic trapping of suspended particulate matter at thesalt wedge interface produces estuarine turbidity maxima (ETM).These ETM events extend particle residence time in the watercolumn, promoting development of specific particle-associatedmicrobial assemblages that support up to ∼80% of the secondaryproduction in the estuary (Simenstad et al., 1990; Crump et al.,1998, 1999).

The activities of particle-attached microorganisms in thedegradation of autochthonous organic matter is speculatedto enhance mineralization of more recalcitrant forms ofallochthonous organic matter, resulting in a “priming effect”with potential to greatly enhance organic matter transformations(Guenet et al., 2010; Bianchi, 2011; Simon et al., 2014). In theColumbia River estuary, the priming effect is associated withtransient algal organic carbon additions, such as perished anddecaying freshwater phytoplankton, the major source of POMfrom spring to fall (Frey et al., 1984; Sullivan et al., 2001; Herfortet al., 2011b). In late summer to early fall, estuarine blooms of themixotrophic ciliate,Mesodinium sp., likely provide an importantautochthonous POM source (Herfort et al., 2011a). In addition,relatively low river discharge allows oceanic water masses to enterthe estuary and to reach 30–40 km into the navigation (South)channel (Simenstad et al., 1984). During upwelling periods, the

tide often carries coastal phytoplankton blooms, another majorsource of POM in the lower estuary (Kudela et al., 2005; Herfortet al., 2011a; Roegner et al., 2011b; Breckenridge et al., 2014).The complex interplay of various POM fluxes is significantbecause our previous results suggested that particle origin is akey determinant of microbial community composition in theColumbia River coastal margin (Smith et al., 2013; Simon et al.,2014).

Previous analyses of mainstem bed of the estuary showed thatsignificant quantities of fine sediment particles originated fromthe Columbia River lateral bays (Sherwood et al., 1984; Simenstadet al., 1984). The tidal-driven flux of nutrients and organic matterfrom such intertidal marshes is called “outwelling” (Odum, 2000),and microbial activity in the sediments and overlying water hasbeen shown to profoundly affect estuarine respiratory gas fluxes(Cai et al., 1999). The lateral bays, therefore, function as hotspotsof organic-matter degradation (Simenstad et al., 1984) deliveringsignificant amounts of reduced chemical species (e.g., methane,ammonium, and dissolvedmetals) to the lower estuary (Cai et al.,1999; Turner andMillward, 2002; Bräuer et al., 2011; Gilbert et al.,2013).

Molecular evidence suggests that the lateral bays alsocontribute to microbial composition in waters of the mainstemestuary. Gene expression and metagenome sequencing studiesrevealed an enrichment of both anaerobic taxa and genes fromanaerobic pathways in oxygenated water samples collected in themainstem estuary, most notably in ETM particles (Smith et al.,2010, 2013; Simon et al., 2014). Those metabolic pathways, whichincluded dissimilatory nitrate reduction, sulfate reduction, andreduction of Mn and Fe oxides (Klinkhammer and McManus,2001; Smith et al., 2010, 2013; Bräuer et al., 2011), were notexpected to occur in the oxygenated water column, and arethought to indicate the presence of anoxic microzones in particles(Smith et al., 2010, 2013; Simon et al., 2014). The associatedtaxa were not enriched in the euphotic coastal ocean samples,suggesting an estuarine origin. We hypothesize that particleerosion from the lateral bays of the lower estuary might providea significant contribution of particulate matter to the estuarinewater column and be responsible for “seeding” the ETM withparticle-attached anaerobic microbes. An initial step towardtesting this hypothesis is characterizing microbial communitiesassociated with the lateral bays.

Our sampling sites were selected to reflect differences in theorigin and character of the sediments, as well as in the degreeof tidal forcing and salt water intrusion that are observed inthese regions. Baker Bay (BB) is a shallow, intertidal inlet locatedbetween river miles 3 and 9, with insignificant local tributaryinflow (Figure 1). It is connected to the North Channel of themain estuary by three conduits, each with dramatic differencesin the phase and magnitude of water exchange (Simenstadet al., 1984). The shallow bay interior (including our samplingsite) experiences maximum salinities well below ocean values(Simenstad et al., 1984). Youngs Bay is another large, shallow,oligohaline lateral bay, with the mouth located between rivermiles 11 and 17 at the confluence of the Youngs and ColumbiaRivers (Figure 1). The Youngs, and Lewis and Clark Riversprovide significant annual local tributary runoff, two-thirds of






FIGURE 1 | Contour map of the lower Columbia River estuary. The shore locations of three sediment sampling sites in Baker Bay and Youngs Bay are shownas black triangles, BB, Baker Bay; YB-M, Youngs Bay mouth, and YB-B, Youngs Bay back. The white triangles show locations of the SATURN endurance stationsproviding continuous flow-through biogeochemical measurements (including salinity, chl a concentrations, etc.).

which occurs in winter, with less than 5% of the total flowoccurring in July–September (Simenstad et al., 1984). Salinities atYoungs Bay mouth (at the YB-M sampling location) range fromfresh under high river flow, to more than 20 PSU under low-flow summer conditions (Simenstad et al., 1984). Historically,drainage of 1000s of acres of tidal marsh and swamp aroundYoungs Bay led to enhanced flushing of organic material, silt, andnutrients by the rivers into the bay (Lev et al., 2008). Compared toBB, Youngs Bay is more organic matter rich, supporting a higherabundance of prey for invertebrates, macroinvertebrates, and fish(Bottom et al., 2005).

In this pilot project, we performed high-throughput Illuminasequencing and metagenome analysis to identify the dominantmicrobial assemblages and pathways involved in carbontransformations in a few selected sediment samples collectedjust below the surface at three sites in lateral bays of thelower estuary (Baker and Youngs bays; Figure 1). Integration ofour sequencing data with continuous, in situ water monitoringobservations from the estuary, plume and coastal ocean togetherwith numerical simulation of water mass transport allowed usto generate testable predictions about the origin and fluxes ofparticles and their impacts on microbial community compositionin the estuary. The study provides important information aboutthe dominant microbial populations and pathways responsiblefor carbon respiration in the estuary and nutrient exchangesalong the river-to-ocean continuum.

Materials and Methods

Sediment Sample Collection andBiogeochemical CharacterizationThree estuarine sediment samples of ∼50 g each were manuallycollected on August 22nd, 2011 near shore during low tide in

three locations, one in BB and two in Youngs Bay (YB-M andYB-B, Youngs Baymouth and back, respectively; Figure 1). Coreswere collected from the wet near-surface sediment (upper 5–10 cm) using sterile 50 mL Corning tubes. Excess water wasdrained from the tubes before homogenizing and aliquotingsediment in 1 g portions. Total DNA isolation was performedusing aliquots stored on ice for 1 day. Frozen (−80◦C) aliquotswere used to obtain chemical characteristics using Agri-Check(Umatilla, OR), and Table 1 shows results of bulk chemicalanalysis for each core.

SATURN Observation Stations and GliderMissionsThe biogeochemical data collected by four observation stationsof the Center for Coastal Margin Observation and Prediction(CMOP) Science and Technology University Research Network(SATURN) were analyzed for ∼1.5 months prior to sedimentsampling. The station locations are shown in Figure 1,each station provided near real-time continuous sensormeasurements, including salinity, temperature, tidal height, andconcentrations of nitrate, dissolved oxygen, and chlorophyll a(chl a) as previously described (Gilbert et al., 2013). The dataare available through the CMOP website (1Baptista et al., 2015).The glider missions were performed using the CMOP Slocum200 m gliders, the instrumentation description, glider tracks,and data are available at http://www.stccmop.org/research/glidermission.

Numerical Simulation of Water Mass TransportThe likelihood of oceanic waters entering Youngs Bay prior tothe sampling time was investigated by running a hydrodynamicalcirculation model (Zhang and Baptista, 2008) for 1 month

1http://www.stccmop.org/datamart/observation_network


http://www.stccmop.org/research/glidermission

http://www.stccmop.org/research/glidermission

http://www.stccmop.org/datamart/observation_network





TABLE 1 | Results of bulk chemical analysis of sediment samples.

Sample name BB YB-M YB-B

Lateral bay Baker Bay Youngs Bay mouth Youngs Bay back

Latitude N46.28551 N46.17469 N46.15968

Longitude W124.05187 W123.84933 W123.80651

Date 22 August 2011 22 August 2011 22 August 2011

Time (PST) 14:50 13:10 16:20

Class Sandy loam Silt loam Silt loam

Salinity (PSU)∗ 4–5 2 0

pH 7.5 6.6 6.3

Organic matter (%) 1.8 5.4 8.8

Total P (ppm) 17 6 15

Total Ca (meq) 7.1 10 11.6

Total Mg (meq) 8.1 8.1 10.8

NO3− (ppm) 2 3 2.5

NH4+ (ppm) 15.5 150 170

Total Mn (ppm) 20 243 162

Total Cu (ppm) 2.4 3.6 3.3

Total Fe (ppm) 130 214 254

∗Water salinity data (PSU) are from measurements collected when river dischargelevels, phase of the tidal cycle, and temperature data corresponded to those of thesampling date.

starting from August 1, 2011. The model was run at thetemporal resolution of 36 s with the model state stored every15 min (Kärnä et al., 2015). The model was spun up for thefirst 7 days, and then the simulation analysis was carried outfor the period of August 8–23, 2011. Three indicator tracerswere added to the simulation to track the fate of riverine,oceanic, and plume water masses. Riverine water mass wasdefined as water entering the estuary from the tidal river(upstream of Puget Island). Oceanic and plume water masseswere defined as water entering from the shelf sea (downstreamfrom the mouth), whose salinity is ≥30 or <30 PSU, respectively.The water age method (Delhez and Deleersnijder, 2002) wasused to compute the time that was spent by these watermasses in the estuary. In this method each indicator tracer isaccompanied by an “age concentration” tracer that accumulatestime proportionally to the indicator tracer. The mean age thatwaters have spent in the estuary can then be computed asthe ratio between the age concentration and indicator tracers.Using the simulation outputs, we computed the percentages ofthe three types of water masses residing in Youngs Bay at anygiven time, as well as their volumetric fluxes into the bay, andmean age.

DNA Preparation and Illumina SequencingGenomic DNA was extracted from 1 g sediment using theFast DNA spin kit for soil (Thermo Fisher Scientific, Waltham,MA, USA). DNA concentrations were quantified using thePicoGreen dye (Invitrogen Corp., Carlsbad, CA, USA), on aNanodrop fluorometer (Thermo Scientific, Wilmington, DE,USA). Approximately 10–30 μg of total DNA was generatedfrom each sample. DNA library preparation and sequencingwere performed at the Oregon Health and Science UniversityMassively Parallel Sequencing Shared Resource (OHSU-MPSSR).

In total three libraries, each containing DNA from a singlesediment sample, were analyzed using Illumina HiSeq 2000Sequencer (Illumina, San Diego, CA, USA), by 2 × 100 bppaired-end sequencing and the base-calling pipeline (IlluminaPipeline-0.3). Sequencing statistics of the processed data are listedin Table 2.

Quantitative PCRThaumarchaeota amoA gene copies were quantified in thesediment DNA using qPCR as previously described (Xu et al.,2012). Standard curves were generated by serial dilutions oflinearized plasmids from sediment samples collected on theWashington bank of the Columbia River estuary (105–101template copies). Melting curves were visually inspected to checkfor a single peak at expected temperature using MyiQ software(v. 1.0.410, BioRad, USA). LinRegPCR software (v. 11.1.0.0,Heart Failure Research Center, The Netherlands) was used tocalculate PCR efficiencies, which were all >1.8. The results werenormalized for starting template concentration.

TABLE 2 | Metagenome statistics.

General BB (BakerBay)

YB-M (YBmouth)

YB-B (YBback)

Number of reads 300,954,858 366,666,206 296,381,839

Individual read length (bp) 101 101 101

Total sequence length (Mbp) 30,400 37,000 30,000

SSU rRNA

Hmmer-selected SSU rRNAreads

55,850 107,241 126,943

Median contig length 219 227 306

Number of selected contigs∗ 927 1241 1111

% assembled of all SSU rRNA 86.7 87 90.1

Mbp assembled into contigs 4.8 9.4 11.5

Average coverage per contig 13.44X 20.11X 25.96X

Metagenome assembly

Total scaffold length (bp) 537,918,774 322,537,715 278,210,246

# scaffolds 507,501 323,378 303,266

Max scaffold length (bp) 220,933 209,382 373,612

% scaffolds containing ≥2ORFs

14.9 15.3 14

Median coverage 7.22X 7.72X 10.04X

Median scaffold length (bp) 611 604 587

Effective bacterial genome size 2.71 2.75 2.54

Number of effective genomes 124 81 74

IMG annotation

Gene count 928,712 576,268 524,872

Gene per scaffold 1.83 1.78 1.73

rRNA count 946 1006 824

Phylogeny, % of gene count 69.5 71.4 73.2

Function, % of gene count 58.6 55.9 57.7

COG, % of gene count 58.07 51.09 56.67

Enzyme, % of gene count 20.19 19.18 21.18

Pfam, % of gene count 71.24 68.11 69.85

∗At least 2X coverage, 5 or more reads, contig length >130 bp.






Pre-assembly Metagenome Analysis of SmallSubunit rRNA and Marker GenesThe taxonomic composition of several marker genes wasidentified and the corresponding taxon abundances quantifiedin the unassembled metagenomes prior to assembly, to avoidthe biasing of abundance metrics by the metagenome assemblypipeline (which is especially pronounced for both rare and veryabundant sequences). Specific marker genes, including smallsubunit rRNA (SILVA rRNA database project (2Quast et al.,2013), methyl coenzyme M reductase (mcrA), and ammoniamonooxygenase (amoA), were searched with the HMMER3software (3Finn et al., 2011) using the probabilistic methods ofprofile hidden Markov models (profile HMMs). The HMMER3-selected reads were then assembled into contigs using Geneious6.1.5 software (Geneious, Auckland, New Zealand). Each contigwas associated with an abundance value expressed as the numberof corresponding unassembled 101-bp reads.

For SSU rRNA, from 25 to 127 1000 101-bp reads permetagenome were selected by HMMER3, 87–90% of which weresubsequently assembled into contigs (Table 2). Contigs wereselected for further analysis if they: (i) exceeded 130 bp in length;(ii) were constructed from ≥5 individual reads in both forwardand reverse orientations; and (iii) had ≥2X coverage and ≥20 bpoverlap between aligned reads. These criteria resulted in ∼1000SSU rRNA contigs per sample, with a mean coverage of 13.4–26X(Table 2). Assembled SSU rRNA contigs were annotated usingthe CAMERA (Community Cyberinfrastructure for AdvancedMicrobial Ecology Research and Analysis, camera.calit2.net)Portal with the rRNA Taxonomy Binning workflow, and SILVAdatabases (Quast et al., 2013). Abundance metrics were preservedfor each assembled contig as the number of initial 101 bp readsused for assembly. Thus, the abundance of a taxon identified bySSU rRNA analysis in a sediment metagenome was calculatedas a percentage of the unassembled reads corresponding tothis taxon over the total number of unassembled SSU rRNAreads.

Two functional marker genes, mcrA and amoA, werealso analyzed using HMMER3 and Geneious 6.1.5 software(Geneious, Auckland, New Zealand). Resulting contigs wereannotated using BLAST searches in CAMERA and at the NCBIBLAST portal (Johnson et al., 2008).

Sequence Data Processing and MetagenomeAssemblyUnassembled reads that passed initial QC were filtered using aquality score cutoff of 30 over 95% of the read length, whichresulted in removal of ∼45% of the total reads. Normalization ofread coverage and read partitioning was performed with khmerv0.6 as outlined in the ‘Partitioning large data sets (50 m+ reads)’protocol (4Pell et al., 2012). Specifically, digital normalization wasperformed with a k-mer size of 20, a cutoff of 10, and four hashtables, while read partitioning was conducted with a k-mer size of32, four hash tables, and a minimum hashsize of 128E+9. Read

2http://www.arb-silva.de/3http://hmmer.janelia.org/4http://lyorn.idyll.org/∼t/khmer/index.html

pairs were retained in bins using a custom Perl script available athttps://github.com/nyoungb2/metagenome_assembly. Each binwas separately assembled into a scaffold using IDBA_UD v1.1.1with default parameters (Peng et al., 2012).

Between 300 and 500 1000 scaffolds were assembled, withmedian scaffold length of ∼600 bp (Table 2). Only 14–15% ofall scaffolds contained two or more coding sequences (CDS).Because of the assembly process, rare sequences represented byunassembled reads not used in scaffolds were excluded fromanalysis. For general peptide analysis, there was additional biasin the abundance metrics from loss of information about thenumber of individual 101-bp reads in the original unassembledmetagenome corresponding to each assembled scaffold. This biaswas most pronounced for very abundant organisms with 100sof 1000s of hits and deep coverage. However, these problemswere outweighed by the improved annotation accuracy achievedwith longer sequences. Thus, we used both assembled andunassembled sequence data iteratively to improve overall resultsfrom taxonomic profiling of the metagenomes.

Analysis of Peptide CompositionThe assembled metagenomes were analyzed using the IntegratedMicrobial Genomes (IMG) with Microbiome Samples – ExpertReview web server (IMG/M-ER of the DOE Joint GenomeInstitute5). The IMG/M-ER server performed prediction of themost probable RNA coding genes, peptide coding sequences(CDS), and subsequent functional and taxonomic annotations(Markowitz et al., 2008, 2012). Peptide prediction was performedby the IMG/M-ER annotation pipeline via homology [basic localalignment search tool (BLAST)] using the E-value cutoff of 10−5

as the default. Functional and phylogenetic profiling was doneusing the IMG collection of reference genomes, and peptidesequence databases of COGs (clusters of orthologous groups),Enzymes, and Pfam (protein families).

Data Submission and Accession NumbersThe three assembled Illumina datasets are available in theIMG/M-ER metagenome database as a part of the Study withthe GOLD ID Gs0047387 (Marine and estuarine microbialcommunities from Columbia River Coastal Margin). Themetagenomes have the following taxon object IDs and projectIDs: (i) BB, 3300001371, Gp0055899; (ii) YB-B, 3300001372,Gp0055898; and (iii) YB-M, 3300001373, Gp0056422.

Metagenome Comparison and EffectiveGenome Size EstimationMetagenomes were analyzed by similarity at taxonomic andfunctional levels. Taxonomic comparison was performed at either(i) the SSU rRNA level using unassembled metagenomes, or (ii)the CDS (predicted peptide) level using assembledmetagenomes.For a particular taxonomic category, the relative abundance valuewas calculated in a metagenome as the sum of rRNA readsor CDS annotated to this taxon divided by the sum total ofall rRNA or CDS predicted for the corresponding metagenome(Table 2). In some cases (indicated in the text), the abundance

5https://img.jgi.doe.gov/cgi-bin/mer/main.cgi


http://www.arb-silva.de/

http://hmmer.janelia.org/

http://lyorn.idyll.org/~t/khmer/index.html

https://github.com/nyoungb2/metagenome_assembly

https://img.jgi.doe.gov/cgi-bin/mer/main.cgi





values for bacterial families were expressed as percentages of totalpredicted bacterial CDS. Thus, the significance of a differencebetween two given samples was calculated based on the twosample t-test between percentage values. Functional comparisonof metagenomes was done using normalized difference, orD-score and D-rank calculated as described (Markowitz et al.,2008 available in IMG/M-ER).

Effective bacterial genome size was calculated using apreviously defined set of 35 core predicted bacterial markerproteins as described (Raes et al., 2007). The size values were usedto calculate the number of effective bacterial genome equivalentsfor each metagenome (74–124, Table 2). Thus, the relativeabundance values for functional gene categories of interest werecalculated as the total number of peptide hits (CDS) for a categoryin a given metagenome divided by the number of correspondingbacterial genome equivalents (Smith et al., 2013).

Results

Sampling Site CharacteristicsSediment samples were collected from three locations withintwo lateral bays of the tidally dominated lower Columbia Riverestuary (Figure 1). Two locations, BB and the mouth of YoungsBay (YB-M), were periodically exposed to oceanic water masses,with near-shore salinities of ≥5 PSU. A third, in the back reachesof Youngs Bay ∼4 km from the mouth (YB-B), representedmudflats in the freshwater interior (Figure 1). General physicaland biogeochemical characteristics varied greatly between thetwo bays. BB is characterized by a sandy loam bottom witha high proportion of coarse gravel, and a low percentage oforganic matter. In contrast, the Youngs Bay mudflats consist ofsilt loam bottom sediments, with YB-M and YB-B sedimentscontaining 3–4X more organic matter relative to BB (Table 1).

At the time of sampling ammonium concentrations were 10 to12X higher in YB-M and YB-B versus BB, with manganese andiron concentrations at 10-8X and 1.6-2X, respectively (Table 1).In general, the Youngs Bay sediment biogeochemistry wasconsistent with the presence of very active microbially drivenorganic matter remineralization (Gilbert et al., 2013).

Biogeochemical Characteristics of the EstuaryPrior to the Sediment SamplingTo contextualize our sequence results, we analyzedbiogeochemical data collected for estuarine end members(tidal freshwater and coastal ocean) for the period of continuousand pronounced coastal upwelling (data not shown) that started∼50 days prior to our sampling date in August 2011. This periodwas associated with persistently high surface chl a concentrationsin the coastal ocean adjacent to the Columbia River mouth (asobserved in near real-time satellite data, Live Access Server ofOcean Watch, The Coast Watch West Coast Regional Node6).Sensor data collected during two CMOP glider missions betweenGrays Harbor and Quinault (June 20–August 6, 2011) indicatedwidespread, upwelling-related hypoxic conditions at depthsbetween 50 and 175 m, with low-oxygen water reaching thesurface at times. In addition, elevated chl a concentrations (upto 50 μg/L) were observed, likely indicative of upwelling-relatedphytoplankton blooms in the euphotic zone (mostly above 20 mdepth; Figure 2).

Continuous observation data collected from July 1 to August22, 2011 by the four SATURN endurance stations (Figure 1)were also analyzed. The SATURN-02 (plume) station measuredabundant chl a in high-salinity water at 1 m depth, in closeproximity to the Columbia River mouth (Figure 3A, top panel).The estuarine SATURN-03 station in the navigation (South)

6http://las.pfeg.noaa.gov/oceanWatch/oceanwatch.php

FIGURE 2 | Glider tracks in the Columbia River coastal margin (two missions between Grays Harbor and Quinault) during the same time periodshowing (i) the upwelling-related hypoxic zone formation (upper plot with depth values colored by dissolved oxygen values) and (ii) phytoplanktonbloom development (lower plot with depth values colored by chl a concentrations). The color scales are shown to the right of each plot. Gaps indicatemissing data.


http://las.pfeg.noaa.gov/oceanWatch/oceanwatch.php





channel showed a clear association between elevated chl aconcentrations and high salinity water at all three measureddepths (2.4, 8.2, and 13 m; the latter two, having higher chl alevels, are shown in Figure 3A). Interestingly, the chl a peaksin the plume and estuary were observed almost simultaneously(with some exceptions early in July, Figure 3A). High salinitywater at SATURN-03 contained high nitrate (Figure 3B) andlow oxygen concentrations typical of upwelled water tidallytransported into the estuary (Roegner et al., 2011a,b). In contrastto the high chl a observed in the estuary-plume continuum,concentrations at the freshwater Columbia River (SATURN-05)station were very low and did not display the prominent peaksindicative of phytoplankton blooms (Figure 3A, bottom panel).

Water Mass Fluxes between the MainstemEstuary and Youngs BayPrevious observations indicated that tidally transported oceanicparticles could only reach the lateral bay sampling locationduring late summer/fall periods with correspondingly low riverdischarges (Chawla et al., 2008). We used numerical simulationof water mass transport during a 2-weeks period prior to thesampling time in August 2011 to evaluate the fluxes, percentagesand mean ages of riverine, plume and oceanic water massesresiding in Youngs Bay at any given time (Figure 4). The resultsindicated that both oceanic and plume water masses regularly

entered the bay, especially during spring tides (August 12–16), when they could occupy up to 18 and 9% of the totalvolume, respectively. This intrusion was weaker during neaptides (August 19–23), with the daily maximum percentagesdecreasing approximately twice (Figure 4). The mean water agein YB indicated that oceanic/plume waters entering the bay wererelatively young during spring tides (less than 40 h after enteringthe estuary), but the small fraction of water residing in the baythroughout the ebb tides can be considerably older (100 h).

Metagenome FeaturesThe initial number of unassembled 101-bp reads was similarfor all three metagenomes, ranging from 300 to 360 million(Table 2). Metagenome assembly reduced the metagenome sizesby approximately two orders of magnitude, in total, from 524to 929 1000 genes predicted for each metagenome (Table 2).Each CDS within a scaffold was considered as a single entry (apredicted peptide) even if it was assembled from multiple reads.We utilized this approach because the majority of scaffolds wereshort (with the average number of genes 1.73–1.83, Table 2), andnot particularly abundant, due to high complexity of the sedimentmicrobiota. During the downstream analysis, each predicted CDSof an assembled scaffold was annotated separately and consideredas a single independent entry (a peptide). Approximately 70 and

FIGURE 3 | (A) Continuous flow-through measurements of chl a concentrations (Y-axis, μg/L) at SATURN endurance stations from July 1 to August 20, 2011. Chl avalues are colored by water salinity (the color scale is shown to the right, in PSU). (B) Continuous flow-through measurements of nitrate (Y-axis, μM) colored bysalinity. Station names and measurement depths are shown in the upper right parts of the panels. Gaps indicate missing data.






FIGURE 4 | Numerical simulations of water mass transport in Youngs Bay during a 2-weeks period prior to the sampling time in August 2011. Theplots show the calculated values for (i) fractions of the total volume residing in Youngs Bay (top), (ii) fluxes into the bay (middle), and (iii) mean age of oceanic,riverine, and plume water masses (bottom).

57% of all predicted peptides were associated with taxonomicand/or functional annotations, respectively (Table 2).

For eachmetagenome, the effective genome size was estimatedusing predicted bacterial peptides (Raes et al., 2007), resulting invalues ranging from 2.5 to 2.75 Mbp (Table 2). These calculatedvalues were likely underestimates (Konstantinidis et al., 2009),since only annotated bacterial peptides were considered andunclassified CDS were not taken into account. Average GCcontent of the metagenomes ranged from 41 to 55% with a single,dominant peak, except for the YB-M sample, which had twodistinct peaks in %GC distribution: one at 41%, and the secondat 61% (data not shown).

Domain Structure of Sediment MetagenomesDomain composition of sediment metagenomes was comparedusing two independent analyses: (i) HMMER3-selected SSUrRNA contigs from unassembled metagenomes, and (ii)predicted peptides (CDS) from assembled metagenomes(Table 3). Comparison of SSU rRNA and peptide annotationsgave almost identical results for the BBmetagenome, with 90% ofsequences belonging to bacteria, archaea constituting 8.7% (SSUrRNA) and 8.9% (peptides), and only 1.4% (SSU rRNA) and 1%(peptides) annotated as Eukaryota (Table 3). In contrast, the SSUrRNA approach indicated much higher proportions of eukaryoticsequences in Youngs Bay metagenomes in comparison with thepredicted peptide approach (7 and 49% of SSU rRNA, versus

TABLE 3 | Domain composition of the three sediment metagenomes.

SSU rRNA Bacteria Archaea Eukaryota Chloroplasts

BB 89.63 8.72 1.40 0.25

YB-M 83.08 0.78 6.75 9.40

YB-B 48.93 1.54 49.08 0.45

Peptide Bacteria Archaea Eukaryota Other

BB 89.90 8.90 1.09 0.11

YB-M 96.95 1.09 1.78 0.18

YB-B 92.56 5.68 1.64 0.11

Abundances are expressed as percentages of total number of either SSU rRNAreads detected by HMMER from unassembled sequence reads (upper part), orpeptide annotations (CDS with ≥30% identity for ≥70% of the alignment length)generated for assembled metagenomes (lower part). Metagenome names indicatesampling locations, BB, Baker Bay; YB-M, Youngs Bay mouth; and YB-B, YoungsBay back.

∼1.8 and 1.6% of predicted peptides in the YB-M and YB-B,respectively; Table 3). The majority of eukaryotic SSU rRNAsequences identified in the YB-B metagenome belonged toGastropoda mollusks (Figure 5). Transient appearance of largenumbers of Gastropoda species (up to 200,000 snails/m2),namely the invasive New Zealand mudsnail, Potamopyrgusantipodarum, was previously observed in Youngs Bay (reviewedin Bersine et al., 2008). The most abundant 18S rRNA contigfrom the YB-B metagenome (38% of all SSU rRNA reads, 2240X






FIGURE 5 | Taxonomic composition of the domain Eukaryota based on SSU rRNA reads detected by HMMER in the unassembled metagenomes. Thepie charts show taxon abundance, in percentage of total Eukaryotic reads, with color legend given below. Metagenome names indicate sampling locations, BB,Baker Bay; YB-M, Youngs Bay mouth; and YB-B, Youngs Bay back. The numbers in brackets represent percentages of Eukaryotic reads in all of the SSU rRNAreads for each metagenome.

coverage) was 100% identical to the reference P. antipodarumSSU rRNA sequence (haplotype 1). The paucity of Eukaryotaobserved at the peptide level in the Youngs Bay metagenomes(Table 3) can be explained by the absence of reference peptidesequence information for large multicellular eukaryotes. Infact, there are currently no sequenced genomes available forGastropoda, leaving only partial rRNA sequences to serve asannotation references. Apparently, most of P. antipodarumpeptide sequences from YB-B remained un-annotated, andtherefore excluded from downstream analyses. As the result,the total number of peptides predicted for the Eukaryota-richYB-B metagenome was almost twice smaller than for theEukaryota-poor BB (Table 2).

EukaryotaAs mentioned above, the YB-B metagenome contained mostlysequences of the Gastropoda P. antipodarum, whereas theYB-M metagenome contained predominantly sequences ofAnnelida (mud worms), Arthropoda (copepods), Magnoliophyta(angiosperms), and Bacillariophyta (diatoms; Figure 5). Diatomnuclear-encoded 18S rRNA sequences constituted ∼0.7% of thetotal SSU rRNA, and most (73%, or 529 hits) were classifiedas highly identical (98.6%) to the brackish water estuarinetaxon Pleurosira laevis. In contrast, the highly abundant diatomchloroplast-encoded 16S rRNA sequences (8489 hits, 9.4% of allSSU rRNA sequences, Table 3) were mostly (97%) attributedto a different taxon, the pelagic marine diatom Odontella(with 99.4% identity over the 1487 bp reference, the Odontellasinensis chloroplast 16S rRNA). The peptide data for diatomsdid not indicate such high abundance (Table 3), most likelybecause the corresponding reference genomes have not yet beensequenced. Nevertheless, the YB-Mmetagenome did contain 165peptide sequences annotated as Bacillaryophyta at the ≥90%level of identity (over at least 70% of the alignment length).Interestingly, nine predicted peptide sequences were annotated

as the large subunit of ribulose 1,5-bisphosphate carboxylase(RuBisCo, pfam00016), in addition, 10 corresponded to thephotosystem I psaA/psaB protein (pfam00223). Both of thesefunctional gene categories, comprising ∼12% of the identifiedhits, are chloroplast-encoded. The majority of RuBisCo peptidesshared 96–99% identities with the reference sequences fromOdontella, Nitzschia, Cocconeis, Stephanodiscus, Cyclotella, andSurirella, with one sharing 100% identity to the P. laevis gene.

BacteriaIn all three metagenomes, at least 90% of classified peptideswere attributed to Bacteria (Table 3). The sediment sequenceswere more evenly distributed across several abundant bacterialtaxonomic groups (Figure 6), instead of being dominatedby Proteobacteria and Bacteroidetes phyla as observed formetagenomes from Columbia River estuarine, plume and coastalocean waters (50 to >85%, Smith et al., 2013). As expected,the sediment metagenomes were also relatively more complex,revealing higher diversity of both predicted SSU rRNA andpeptides. The most striking differences in bacterial phyla amongthe lateral bay metagenomes were (i) a high abundance ofChloroflexi (in particular, Anaerolineaceae) in BB (14% of thetotal); (ii) relatively high numbers of Verrucomicrobia in bothYB-M and YB-B metagenomes (3–4%); (iii) a predominanceof Bacteroidetes in YB-M (58%); and (iv) a relatively highproportion of Acidobacteria in YB-B (7%; Figure 6).

In-depth taxonomic analysis of the phylum Bacteroidetesdominating the YB-M metagenome (58 and 28% of predictedpeptides and SSU rRNA, respectively, Figure 6) showed higherrelative abundance of Flavobacteria (6X), Cytophagia (5X),and Bacteroidia (10X) compared with metagenomes from theother two locations [at both peptide (Figure 7A) and SSUrRNA levels; data not shown]. Flavobacteria are known todegrade polysaccharides associated with phytoplankton particles,and therefore contain genes for carbohydrate-active enzymes






FIGURE 6 | Phylogenetic composition of the domain Bacteria. For eachmetagenome, the left bar (R) represents taxon abundance based on SSUrRNA composition expressed as the percentage of total number of bacterialSSU rRNA reads detected by HMMER in each unassembled metagenome.The right bar (P) represents taxon abundance based on peptide annotations(CDS with ≥30% identity for ≥70% of the alignment length). Each valuerepresents the percentage of all corresponding CDS relative to the totalnumber of bacterial CDS in an assembled metagenome. Metagenome namesindicate sampling locations, BB, Baker Bay; YB-M, Youngs Bay mouth; andYB-B, Youngs Bay back.

(CAZymes, EC:3.2.1.-Hydrolases. Glycosylases. Glycosidases;Teeling et al., 2012). Thus, we analyzed the relative abundance ofthe CAZyme functional gene group, normalized by the number ofeffective bacterial genomes in eachmetagenome. A CAZyme genewas selected for analysis if it corresponded to ≥100 predictedpeptide hits within a metagenome. We found 21 abundantCAZyme genes (selected from 53 CAZyme genes identified inthe sediment metagenomes), 14 of which were enriched ≥2-fold in the YB-M compared to BB and YB-B metagenomes(Figure 7B). This enrichment was especially pronounced forfucose permease and α-L-fucosidase (4.6 and 3.4X versus BB,respectively), both of which are associated with utilizationof fucose, a major component of diatom exopolysaccharides(Teeling et al., 2012).

The freshwater YB-B sediment contained abundantBetaproteobacteria (15% of predicted bacterial peptides). Furtheranalysis at ≥60% sequence identity (roughly correspondingto family/genus) showed that approximately one half of thecorresponding sequences represented various taxa from theorder Burkholderiales, whereas another quarter correspondedto the family Gallionellaceae (data not shown). In contrast, thetwo sediments from locations exposed to oceanic water masses,YB-M and BB, were dominated by Deltaproteobacteria, whichconstituted from 12 (YB-M) to 26% (BB) of all predicted bacterialpeptides (at ≥30% sequence identity). At ≥60% identity level, upto 45% of all Deltaproteobacteria sequences belonged to severaldifferent orders of dissimilatory sulfate-reducing bacteria (SRB;Figure 8A), with highest abundance observed for the familiesDesulfobacteraceae andDesulfobulbaceae – also prevalent in SRB

FIGURE 7 | (A) Abundance of the four Bacteroidetes classes at the peptide level expressed as the percentage of all corresponding CDS relative to the total numberof bacterial CDS in an assembled metagenome. (B) Carbohydrate-active enzymes (CAZymes, EC:3.2.1.-Hydrolases. Glycosylases. Glycosidases) showing 14 out of21 CAZymes with >100 peptide hits that were enriched in YB-M relative to BB and YB-B. Abundance for a functional gene category in an assembled metagenomewas calculated as the number of corresponding CDS, normalized by the effective bacterial genome equivalents in a metagenome. Two genes specifically associatedwith the degradation of diatom cell wall polysaccharides are boxed.






FIGURE 8 | (A) Abundance of sulfate reducing bacteria (SRB, bar graph) andthe associated marker gene dsr (desulfoviridin, bubble plot). Abundancevalues for SRB were calculated based on peptide annotations at thefamily/genus level (≥60% identity for ≥70% of the alignment length). Eachvalue represents the percentage of all corresponding CDS relative to the totalnumber of bacterial CDS in an assembled metagenome. Abundance for thedsr gene was calculated as the corresponding number of hits, normalized bythe effective bacterial genome equivalents in the assembled metagenome.The values are shown with bubble width, the scale is shown as a half-bubbleto the right (the number shows the scale value, two genes per the effectivebacterial genome equivalent). Metagenome names indicate samplinglocations, BB, Baker Bay; YB-M, Youngs Bay mouth, and YB-B, Youngs Bayback. (B) Enrichment of Geobacteraceae and Pelobacteraceae (Mn and Fereducing bacteria). The abundance values were calculated as describedabove.

populations of other brackish to marine estuarine environments(Bahr et al., 2005; O’Sullivan et al., 2013). Consistent with theirtaxonomic abundance, the marker genes for dissimilatory sulfitereduction, encoding subunits of desulfoviridin (dsr), were alsoenriched in the BB and YB-M compared to YB-B sediments(3.5 and 1.7X, respectively, the bubble plot in Figure 8A). Incontrast to SRB, other Deltaproteobacteria, namely the familiesGeobacteraceae and Pelobacteraceae representing the Mn/Fe-reducing anaerobic bacteria, had the highest abundance in YB-M(Figure 8B).

ArchaeaThe highest abundance of Archaea, constituting ∼9% of allSSU rRNA and predicted peptides, was observed in the BBmetagenome (Table 3). In YB-M and YB-B, respectively, 0.8and 1.5% of the archaeal sequences were identified from SSUrRNA analysis, and 1–5.7% from predicted peptides (Table 3).Differences in results from these approaches likely reflect anumber of issues including: (i) variation in copy numbers ofrRNA genes, (ii) relative scarcity of annotated peptide referencesavailable for Archaea, and (iii) annotation of archaeal genes asbacterial due to sequence similarities (Lloyd et al., 2013). Somearchaeal groups, such as Thaumarchaeota andMethanomicrobia,showed consistency between the two approaches (Figure 9). Incontrast, Methanobacteria were observed mainly from predictedpeptide data, and the “Miscellaneous Crenarchaeotal Group”(Kubo et al., 2012) was identified only in SSU rRNA data(Figure 9). To classify the predominant Archaea present in thesediment metagenomes, we analyzed the most abundant archaealSSU rRNA sequences represented by long contigs (using selection

FIGURE 9 | Phylogenetic composition of the domain Archaea. For each metagenome, the left bar (R) represents taxon abundance based on SSU rRNAcomposition expressed as the percentage of total number of archaeal SSU rRNA reads detected by HMMER in each unassembled metagenome. The right bar (P)represents taxon abundance based on peptide annotations (CDS with ≥30% identity for ≥70% of the alignment length). Each value represents the percentage of allcorresponding CDS relative to the total number of archaeal CDS in an assembled metagenome. Metagenome names indicate sampling locations, BB, Baker Bay;YB-M, Youngs Bay mouth; and YB-B, Youngs Bay back.






criteria of ≥4X coverage and ≥1 Kb length; Figure 10). Thehighest number of contigs (6) meeting these criteria was observedin the BB metagenome, followed by YB-M (3) and YB-B (1).

ThaumarchaeotaOne relatively long contig from YB-B (YB-B-12 in Figure 10)was 97% identical (over 1477 bp) to the 16S rRNA gene from

the freshwater thaumarchaeon Nitrosoarchaeum limnia. Readcoverage for this contig was high (average of 65X coverageover the full length). Because N. limnia is a member of theammonia-oxidizing archaea, we performed a HMMER3 searchof the unassembled YB-B metagenome for reads correspondingto the functional gene for ammonia monooxygenase subunit A(amoA). Although only one full length (590 bp) amoA contig

FIGURE 10 | Inferred phylogenetic tree showing archaeal 16S rRNA contigs from the lateral bay sediment metagenomes (UPGMA, bootstrap analysiswith 100 replicates). Only contigs with at least 4X coverage that were >1 kb in length are shown. Gray circles (BB), gray diamonds (YB-M), and open diamonds(YB-B) indicate contigs from Baker Bay, Youngs Bay mouth, and Youngs Bay back, respectively. Phylogenetic clades were anchored using both rRNA sequencesfrom fully sequenced archaeal genomes (shown with organism names) and selected environmental rRNA clones (most closely related to our sequences and shownwith GenBank accession numbers). Numbers represent bootstrap values corresponding to each branch.






was annotated from the YB-B metagenome, it was 99% identicalto several environmental amoA sequences recovered from ricepaddies and soil, and 95.5 and 91.7% identitical to the N. limniaand N. maritimus amoA sequences, respectively. Furthermore,read coverage was 70X for this contig, which was very similarto that observed for the N. limnia-like 16S rRNA gene. Sinceboth 16S rRNA and amoA genes are found in single-copy in theN. limnia genome, these data suggest that they were recoveredfrom the same abundant representative of an N. limnia-likemember of the Thaumarchaeota in YB freshwater sediments.

Thaumarchaeal amoA genes were also observed, in lowerabundance, in YB-M and BB sediment metagenomes, with 4.1and 4.6X coverage, respectively. These sediments are exposedto higher salinity water that does not penetrate into thefreshwater regions further back in YB. This difference wasreflected by the identity of sequences recovered from the threemetagenomes. In YB-M, a 16S rRNA contig of 1165 bp (YB-M-206 in Figure 10) was 99 and 98% identical to 16S rRNAgenes from Candidatus “Nitrosoarchaeum koreensis” and Ca.“Nitrosopumilus koreensis,” respectively. An amoA contig witha similar coverage (5X) was 93% identical to the N. maritimusgene. Similarly, the BB metagenome contained short 16S rRNAand amoA gene contigs with 93 and 95% identities to thecorresponding genes from N. maritimus.

We used quantitative PCR (qPCR) to determine theabundance of the thaumarchaeal amoA gene in the sedimentcore samples used for metagenome analysis. The results indicatedthat the YB-B sample contained 6.49 × 107 (SD 1.53 × 106)amoA copies per gram of sediment, which was 29 and 18X higherthan amoA copies in YB-M and BB samples, respectively. TheqPCR results were in good agreement with the results fromHMMER3 searches for SSU rRNA and amoA genes indicatingthat Thaumarchaeota populations were approximately 20X moreabundant in the freshwater YB-B metagenome, compared tothose from the more marine-influenced YB-M and BB samples.

Novel and Ambiguous Archaeal TaxaThaumarchaeota represented 69 and 73% of the archaealsequences identified in the YB-B metagenome as 16S rRNAgenes and predicted peptides, respectively (Figure 9). Incontrast, archaeal sequences identified in the BB and YB-M metagenomes were more diverse (Figure 9), with manycorresponding to poorly characterized taxa without fullysequenced reference genomes. One group was classified byIMG/M-ER as Halobacteria (according to SSU rRNA gene andpredicted peptide analyses, Figure 9). However, similarity-basedsearches indicated that the corresponding 16S rRNA genes sharedthe highest identity (98–99%) to PCR-amplified sequences fromestuarine sediments (Abreu et al., 2001), and were only distantlyrelated (82% identity) to well-characterized and fully sequencedHalobacteria from high-salt habitats. Another sequence groupwas classified as Thermoprotei; however, in contrast towell-characterized thermophilic representatives of this taxon,these sequences represented uncharacterized mesophilic archaea(“Thermoprotei-like Crenarchaeota” in Figure 9). One of themost poorly characterized groups, “Miscellaneous CrenarchaeotaGroup” (MCG), was also very abundant (representing ≥10% of

archaeal 16S rRNA sequences) in YB-M and BB metagenomes(Figure 9). Five 16S rRNA gene contigs >1 Kb were identifiedfrom YB-M and BB metagenomes. Of these, two were classifiedas Thermoprotei-like, two as MCG, and one (discussed below)was related to sequences from the Methanomicrobia (Figure 10).

Another poorly characterized, but intriguing group identifiedby SSU rRNA gene sequences in BB and YB-M metagenomeswas the so-called “7th order of methanogens” (Euryarchaeota),related to the Thermoplasmata and recently re-classified as“Methanoplasmatales” (Paul et al., 2012; Borrel et al., 2013). Thisgroup constituted >20% of all identified archaeal sequences inthe BB metagenome (Figure 9), and is represented by two 16SrRNA contigs >1 Kb (with 34 and 14X coverage for BB-17 andBB-43, respectively, Figure 10) that placed phylogenetically with16S rRNA gene sequences of characterized Methanoplasmatales.A long contig (with approximately 7X coverage) correspondingto mcrA, a specific biomarker for methanogenesis encodingmethyl coenzyme M reductase A (Paul et al., 2012), was alsofound in the BB metagenome. This contig was 93% identicalto the mcrA gene of a recently sequenced Methanoplasmatalesrepresentative, Methanomassiliicoccus luminyensis, isolated fromhuman feces (Dridi et al., 2012). The difference in coveragebetween the 16S rRNA andmcrA contigs indicated that additionalsequences of this novel group may have been present, but missedby our searches.

Methanogenic and Methanotrophic ArchaeaBB and YB-M metagenomes both contained an abundance ofdiverse sequences representing well-characterized methanogenicarchaea, including Methanomicrobia and Methanobacteria,in addition to the Methanoplasmatales described above(Figure 9). The BBmetagenome in particular, with approximatelysixfold higher abundance of archaeal sequences, contained avariety of short (200–700 bp) 16S rRNA and mcrA contigswith coverage ranging from 2 to 9X and 86 to 99.5%identity to the corresponding genes from fully sequencedgenomes of methanogenic archaea (i.e., Methanomicrobiaceae,Methanoregulaceae, Methanosaetaceae, andMethanosarcinaceaefamilies).

One long (>1 Kb) 16S rRNA gene contig in the BBmetagenome (BB-63 in Figure 10) related to Methanomicrobiawas 93% identical to sequences from anaerobic methanotrophicarchaea of the ANME-1 clade, which are capable of the anaerobicoxidation of methane (AOM; Meyerdierks et al., 2005). Thesequence with the highest identity (97%) to the BB contig wasrecovered from the Yangtze River mudflats (Zeleke et al., 2013).A HMMER3 search also identified two mcrA contigs with closestidentity (86–90%) to ANME-1 clones from a Mediterraneansubmarine mud volcano (Lazar et al., 2012). Since coverage wassimilar for both the 16S rRNA and the mcrA gene contigs (7.3and 5.9X, respectively), it is possible that they represent the sameANME-1 group in the BB sediment.

Methanogenesis-Related Functional GeneCategoriesMany of the methanogenic archaea observed in the lateral baysamples corresponded to poorly characterized taxa. We therefore






analyzed predicted peptide composition using functional geneclassifications to identify metagenome metabolic properties.The functional categories involved in archaeal methanogenesis(Figure 11) were selected using the MetaCyc Pathway database(Caspi et al., 2012). In total, 11 key enzymes were highlyenriched in the BB metagenome: on average 10 and 6.5X incomparison with YB-M and YB-B metagenomes, respectively(Figure 11). They consisted of the five key enzymes catalyzingmethanogenesis fromH2/CO2 (Figure 11, left panel), plus (i) twokey enzymes of acetoclastic methanogenesis (methyltransferasesMt-cdh, C and D, EC: 2.3.1.169 and EC:2.1.1.245); (ii)the marker gene for methanogenesis from trimethylamine(trimethylamine methyltransferase Mb-mttB, EC:2.1.1.250); (iii)the gene catalyzing cofactor regeneration during methanogenesis(hdr, CoB-CoM heterodisulfide reductase, various subunitsclassified as EC:1.8.98.1) and (iv) two genes involved in the lastcommon steps of methane production in the different pathways:coenzyme M methylation and reduction to methane (H4mpt,tetrahydromethanopterin S-methyltransferase EC:2.1.1.86, andmcrA, co-B sulfoethylthiotransferase; EC:2.8.4.1, pfam02249,02745; Figure 11). Applying the D-rank approach (Markowitz

et al., 2008), we determined that enrichment for 701 functionalgene categories was significant in BB versus both YB-M and YB-B metagenomes (data not shown). Among these categories, thehighest enrichment was observed for the methanogenesis-relatedhdr gene (29 and 36X, p < 0.009 for BB over YB-M and YB-B,respectively). In total, 6 out of 11 methanogenesis-related genecategories shown in Figure 11 were significantly (>2.33, withp-values < 0.009) enriched in BB versus YB-M and YB-B.

Methanotrophic and Syntrophic BacteriaTwo of the identified marker genes [Figure 11; hdr (EC:1.8.98.1),and mch (EC:3.5.4.27)] are involved in both production anddegradation of methane (Chistoserdova et al., 2004). This mayexplain the high numbers of the corresponding sequence hitsobserved, e.g., 2672 for hdr in the BB metagenome. In addition,several other markers of methylotrophy (Kalyuzhnaya et al.,2008) were enriched in the BB metagenome, with normalizedabundances 10–33X higher than those in the YB-M metagenome(Figure 12A).

Enrichment of methylotrophic taxa was also detected inthe YB-M metagenome from analysis of both SSU rRNA

FIGURE 11 | Methanogenesis pathways (methanogenesis from CO2, acetate, and trimethylamine adapted from MetaCyc Pathways), and normalizedabundance of the corresponding genes in sediment metagenomes. The numbers represent enzymes based on the EC classification. The categories EC1.12.98.2 and 1.5.99.11 have been recently transferred to EC 1.5.98.1 and 1.5.98.2, respectively. Abundance for a functional gene category was calculated as thenumber of hits to a given category normalized by the effective bacterial genome equivalents in the corresponding assembled metagenome. Abundance values areindicated by bubble width, with the scale shown as a half-bubble to the right in each panel (the number to the right shows the scale value, from 0.1 to 10 genes perthe effective bacterial genome equivalent). Metagenome names are shown at the bottom left, they indicate the sampling locations: BB, Baker Bay; YB-M, YoungsBay mouth; and YB-B, Youngs Bay back.






FIGURE 12 | (A) Functional gene categories involved in methanotrophy/methylotrophy. The genes are mtdB, coenzyme F420-dependent N(5),N(10)-methenyltetra-hydromethanopterin dehydrogenase; fhcD, formylmethanofuran:tetrahydromethanopterin formyltransferase; fae, formaldehyde-activating enzyme. Abundance of afunctional gene category was calculated as the number of hits to a given category normalized by the effective bacterial genome equivalents in the correspondingassembled metagenome. Abundance values are indicated by bubble width, with the scale shown as a half-bubble to the right in each panel. (B) Relativephylogenetic abundance of putative methylotrophs and synthrophic bacteria are shown as the percentages of all SSU rRNA reads (selected by HMMER3 fromunassembled metagenomes). Metagenome names are shown at the bottom, they indicate the sampling locations: BB, Baker Bay; YB-M, Youngs Bay mouth; andYB-B, Youngs Bay back.

and peptides; most notably for the family Methylococcaceae(Figure 12B). However, a much greater enrichment was observedfor putative Planctomycete methylotrophs (up to 8% of the totalbacterial SSU rRNA reads) in the BB metagenome (Figure 12B).Certain members of the Planctomycetes are believed to carry outmethano- and methylotrophy because they contain genes for C1transfers mediated by methanopterin and methanofuran, similarto Proteobacteria methylotrophs (Chistoserdova et al., 2004).This result, together with the finding of abundant sequencesfrom methane biosynthesis pathways, indicates that the methaneproduced in the sediments at the BB site was at least partiallyconsumed there.

Enrichment was also seen in our data set for two majorgroups of syntrophic and semi-syntrophic organisms from theorder Syntrophobacterales (Deltaproteobacteria; Narihiro andSekiguchi, 2007) and the Anaerolineaceae family (Chloroflexi;Narihiro et al., 2012), respectively. Based on peptide data, theSyntrophobacterales order was enriched approximately 6.5X inthe BB metagenome in comparison with YB-M metagenome.The Anaerolineaceae family represented 5.5% of all bacterialpeptides in the BB metagenome, and was enriched 2.5X incomparison with the YB-M metagenome. Enrichment of bothgroups in the BB relative to the YB-M metagenome waseven more apparent from the SSU rRNA data (17X, and3X for Syntrophobacterales and Anaerolineaceae, respectively;Figure 12).

Discussion

Methanogenesis in Lateral Bay SedimentsThe Columbia River and its major tributary, the WillametteRiver, have dissolved methane concentrations 1–2 orders of

magnitude higher than open ocean waters (Anthony et al., 2012).Additional estuarine sources of methane have been identifiedat intermediate to high salinities, and they include intertidalmudflats (Middelburg et al., 2002). In this habitat, much ofthe organic matter degradation is carried out anaerobically byinteracting groups of methanogenic archaea and SRB (O’Sullivanet al., 2013; Zeleke et al., 2013). Complementary analysisof our sediment samples by Illumina paired-end sequencingand de novo genome assembly of 56 Methanosarcina mazeiisolates revealed phenotypically distinct clades of M. mazeiwith different methane production rates during growth ontrimethylamine (Youngblut et al., 2015). This observation isconsistent with the maintenance of diversity in lateral baysediments through ecological differentiation. Our comparison oflateral bay metagenomes likewise found that SSU rRNA genes ofmethanogens (Figure 9), and predicted peptides from functionalgene categories required for producing methane from CO2,acetate and trimethylamine, were highly enriched in the BBmetagenome (Figure 11). In fact, methanogenesis was the mostobvious microbial metabolism distinguishing BB from YB-M andYB-B metagenomes (3–30X enrichment of the correspondingfunctional gene categories observed in BB). Methanogenicarchaea were also found to dominate the near-surface brackishestuarine sediments of the Yangtze River in China (Zelekeet al., 2013). Interestingly, there were also abundant (butpoorly characterized) putative methylo/methanotrophs fromthe Planctomycetes phylum in the BB metagenome. Our datasuggest that both methanogenesis and methanotrophy occurin BB sediments. Many of the genes encoding componentsof the methanogenic pathway have been found in genomesof bacteria and archaea involved in aerobic and anaerobicmethanotrophy (Chistoserdova et al., 1998, 2004). Methaneproduced in this environment may be oxidized by anaerobic






methanotrophs (ANME) using sulfate as electron acceptor(Knittel and Boetius, 2009), or diffuse across the oxic/anoxicinterface to be oxidized aerobically. Alternatively, methaneproduced in the lateral bay sediments may be advected into themain channel during the ebb stages of the tidal cycle. Previousstudies have shown that gasses are transported from intertidalmarshes into estuaries with the ebb tide, and may surpass theefflux of dissolved organic matter (Cai et al., 1999). Whilenet methane emission into the atmosphere from the ColumbiaRiver estuary seems likely (Middelburg et al., 2002), methanemeasurements indicate that biological consumption does occur(Sansone et al., 1999).

Methanogenesis and Syntrophic BacteriaSulfate-reducing bacteria are a diverse group of obligate,anaerobic bacteria that use sulfate as the terminal electronacceptor during the oxidation of a variety of electron donors,many of them organic (for review see Plugge et al., 2010). SRBare known to outcompete methanogenic archaea for substratessuch as H2 and acetate; thus, high sulfate concentrations inocean tidal water influx favor sulfate reduction as the majorprocess for anaerobic organic matter utilization (Oremland andPolcin, 1982; Leloup et al., 2009). In our work, SRB and theobligate hydrogenotrophic methanogens (i.e., Methanomicrobia)were found in the samemetagenome. One possible explanation isthe fine-scale vertical separation of these two groups in sedimentswas overlooked when the upper 6 cm were collected as a singlesample.

Other methanogens may co-exist with SRB because the latterdo not compete for methanogenic substrates such as methanol,trimethylamine, or methionine (Oremland and Polcin, 1982;Leloup et al., 2009). Moreover, some SRB, i.e., Desulfovibriovulgaris, are capable of metabolic shifting between two lifestyles:from sulfidogenic growth to syntrophic growth in associationwith the methanogen Methanosarcina barkeri (Plugge et al.,2010). In the presence of sulfate, SRB carry out sulfidogenesisand oxidize the products of primary fermentations to CO2.Once sulfate is depleted, they ferment organic acids andalcohols, producing hydrogen, acetate and carbon dioxide,and subsequently rely on hydrogen- and acetate-scavengingmethanogens to convert these compounds to methane (Pluggeet al., 2010). This syntrophy is widespread in marine andestuarine environments, since it allows SRB to survive upondepletion of sulfate (which occurs every tidal cycle; Leloup et al.,2009; Plugge et al., 2010). SRB have been found to be abundant inzones with high methane production (Leloup et al., 2009).

Both YB-M and BB sediments are exposed to elevated sulfateconcentrations periodically with the tides. Consistent with this,abundant sequences representing SRB communities were presentat both locations (with the relative proportion being higher inBB than in YB-M, Figure 9). However, only in BB sedimentsdid the SRB appear to co-exist with an abundant and diversecommunity of methanogenic archaea. Even though the YoungsBay (YB-M) sediments have 3X more organic matter than BB(Table 2), it may be that the sources and specific types oforganic matter present are more important. For example, itwas noted (Macdonald and Winfield, 1984) that clumps of

the marine brown alga Fucus distichus edentatus, an inhabitantof intertidal zones on rocky shores, commonly co-occurredwithin stands of sedges (Carex lyngbyei) in BB only. These andother seaweeds are noted for the relatively high amounts oftrimethylamines associated with both live and decaying cells(Kneifel, 1979). Recent evidence points to obligate, H2-dependentmethylotrophic methanogenesis and consumption of a diversityof methylated compounds bymembers of the Thermoplasmatales(Borrel et al., 2013), a particularly abundant group in the BBmetagenome. Thus, the presence of these algae may play a role inthe selective enrichment for methylotrophic methanogens in BBsediments. The BB metagenome was also enriched for syntrophicand semi-syntrophic bacteria from the order Syntrophobacterales(Narihiro and Sekiguchi, 2007) and the family Anaerolineaceae(Narihiro et al., 2012), which are known to be involved inanaerobic degradation of fatty acids and other carbohydratesand generation of substrates for methanogenesis (reviewed inNarihiro et al., 2012).

Evidence for the Erosion of Lateral BaySediments and Trapping by ETMMany of the same taxonomic groups found in the lateral baysediment metagenomes in this study were previously foundenriched in the large particulate fraction of ETM water samples,relative to the free-living, smaller-size fraction (Smith et al.,2013). These strict and facultative anaerobes included archaealmethanogens together with syntrophic bacterial taxa (i.e., SRBand the Anaerolineaceae family); and the Mn/Fe-reducinggenera, Pelobacter and Geobacter. Unfortunately, the similarity-based peptide annotations used for taxonomic analyses couldnot unequivocally determine whether the ETM and lateralbay sediments were populated by the same organisms (at thespecies level), or whether sequences recovered from these twohabitats represented related, but distinct members of the samefamily and/or genus. In addition, because our analysis wasnot carried out on full-length sequences, but rather on shorterfragments that were below a 97% identity threshold and sequencecomplexity was very high, it was not possible to determinewhether SSU rRNA operational taxonomic units (OTUs) presentin the water and in the sediments represented the sametaxa. Nevertheless, the presence of taxonomically overlappingmicrobiota in ETM particles and lateral bay sediments mayindicate particle transport and ‘seeding’ of the water withfacultative and strict anaerobes from the sediments. This particleexchange would be driven by tidal forces, and possibly enhancedduring spring tides and at times of low river discharge, suchas during late summer/fall. Altogether, these data suggestpotential erosion of lateral bay sediments, followed by theirretention in the estuary in ETM, effectively enriching anaerobicmicroorganisms in the oxygenated water through the existenceof particle-associated anoxic microzones (Smith et al., 2010,2013).

Particle Deposition from the Mainstem Estuaryinto Lateral BaysOur simulation analysis indicated that the water massesoriginating from the continental shelf and plume regularly






entered Youngs Bay in relatively high volumes. Thus, particulatematter, when present, may have settled onto the sedimentsnear the mouth of Youngs Bay during slack tides. Analysisof microbial community composition in fact suggested large-scale deposition of several diatom taxa, including the pelagicmarine Odontella, and brackish water P. laevis. Previousobservations suggested a shift of phytoplankton assemblagestoward marine species under late summer conditions oflow river flow, particularly near the mouth of the estuary(Breckenridge et al., 2014). Consistent with our sequence data,several prolonged periods of coastal ocean upwelling withassociated phytoplankton bloom development were observedfor ∼1.5 months prior to our sample collections (Figure 2).Continuous SATURN station observations showed that high-salinity and high-nitrate water masses from the upwelling-influenced coastal ocean were periodically observed in the mainSouth Channel of the Columbia River estuary in July–August2011, transporting elevated concentrations of chl a (Figure 3A).At the same time, blooms of Mesodinium sp. were not detectedduring August 2011. Freshwater chl a concentrations were alsoconsistently low (Figure 3A, bottom panel), indicating thatthe phytoplankton detected in our YB-M sediment sampleoriginated in marine waters. Together with our sequence data,these results suggest that marine diatoms were transportedinto the estuary with the tides, and deposited as perishedand decaying cells in the bottom sediments near the YBmouth.

Previous microscopic surveys of shallow lateral bay sediments(Frey et al., 1984; Simenstad et al., 1984) revealed assemblagesof mostly benthic diatoms, and did not find marine planktonictaxa in the upper layers of sediment, even in the samplesfrom locations adjacent to the Columbia River mouth. Incontrast, multiple marine diatoms (i.e., Asterionellopsis glacialis)have been identified by microcopy in the estuarine watercolumn at the mouth of Columbia River (Breckenridge et al.,2014). Possibly, prior to sediment deposition, live diatomcells were lysed at the estuarine freshwater-seawater interface(Sherwood et al., 1984), with rapid deterioration of the thinsilica frustules used to distinguish these species under themicroscope (Simenstad et al., 1984). Our data provide indirectevidence of diatom degradation, since diatom chloroplast rRNAgenes, and sequences corresponding to the chloroplast-encodedphotosystem I psaA/psaB and RuBisCo large subunit proteinswere abundant, whereas the nuclear-encoded RuBisCo smallsubunit sequences were not found. Protection of multiplechloroplast membranes is proposed to make chloroplast DNAmore stable in the environment than nuclear DNA uponcell death (Bedoshvili et al., 2009; Smith et al., 2013).Thus, detrital deposition of phytoplankton observable at thechloroplast DNA level could pass undetected by microscopicanalysis.

Interestingly, the deposition of decaying bloom particleshad a dramatic effect on the sediment bacterial communitycomposition, apparently leading to significant enrichment ofBacteroidetes (>50% of all identified bacterial peptides) in theYB-M sample (Figures 6 and 8A). Bacteroidetes/Flavobacteriaare known to be abundant during periods of high primary

production (Williams et al., 2012), and members of thesetaxa degrade phytoplankton polysaccharides (Kirchman,2002). For example, in the North Sea their abundance wasincreased fivefold upon commencement of algal blooms(Teeling et al., 2012). In our previous work we also observedenrichment of Flavobacteria in diatom-replete euphoticzone water samples, compared to samples from the deepocean containing few diatom sequences (Smith et al., 2013).Bacteroidetes genes coding for carbohydrate-active enzymes(CAZymes) have been linked to particle utilization (Teelinget al., 2012). Our analysis of CAZyme functional categoriesin the sediment metagenomes showed a dramatic enrichment(>70%) of abundant CAZyme genes in the YB-M metagenome(Figure 7B). It may be that the Bacteroidetes/Flavobacteriain the YB-M sample were indigenous to the sediments, withtheir growth and activity selected for by deposition events.Alternatively, they may have colonized the diatom biomassin the coastal ocean or the estuary during transport, andwere deposited onto the bottom sediments along with detritalparticles. Thus, during periods of high primary production(such as late summer) lateral bay sediments of the lowerColumbia River estuary may serve as repositories of biogenicparticles, and the allochthonous particle influxes may providea major force shaping microbial community composition andactivities.

Conclusion

Our metagenomic data revealed BB as a potentially importantsite for methanogenesis in the Columbia River estuary. Thisputative methylotrophic methanogen community may play amajor role in distinguishing the microbial biogeochemistryof BB from similarly salinity-influenced regions of YoungsBay.

Our data additionally provide support for reciprocalparticle exchange between the lateral bays and the mainstemwater column. These exchange events can dramatically alterthe microbial community composition and activities, forexample enriching specifically for diatom-degrading bacteriain localized “hotspots” within lateral bay sediments. Severallines of evidence also suggested that lateral bay sedimentsmay be involved in particle “seeding” of ETM, resulting inenrichment of anaerobic microorganisms in the oxygenatedwater column (Crump et al., 1999; Smith et al., 2010,2013; Bräuer et al., 2011; Simon et al., 2014). Analysisof the lateral bay sediment metagenomes supports theseconclusions, revealing related taxonomic groups. Currentlywe are expanding the pilot study through metagenomeanalysis of eight additional sediment samples from Baker,Youngs and Cathlamet Bays of the lower estuary. Thisshould increase our ability to interpret complex abundancepatterns observed for bacterial taxa, to relate specific microbialmetabolisms to biogeochemical measurements, and ultimately,to better define the roles of lateral bays in the generationof net ecosystem metabolism in the lower Columbia Riverestuary.






Author Contributions

MS, LH, and HS conceived and designed the study, collected theenvironmental samples, performed the experiments, conducteddata analysis, and wrote the manuscript. RD and NY performedmetagenome assembly, annotation, and comparative dataanalyses. TK and AB performed numerical simulations of watermass transport. RW, WM, and BT contributed to the dataanalysis, manuscript writing and editing.

Acknowledgments

We gratefully acknowledge the CMOP Modeling andCyberinfrastructure teams for providing access to outputs fromthe SELFE model and environmental data collected with aSlocum 200 m underwater glider and endurance stations ofthe SATURN observation network. Support was provided byNational Science Foundation grants (OCE 0424602 and MCB0644468).

References

Abreu, C., Jurgens, G., De Marco, P., Saano, A., and Bordalo, A. A. (2001).Crenarchaeota and Euryarchaeota in temperate estuarine sediments. J. Appl.Microbiol. 90, 713–718. doi: 10.1046/j.1365-2672.2001.01297.x

Anthony, S. E., Prahl, F. G., and Peterson, T. A. (2012). Methane dynamicsin the Willamette River. Oregon. Limnol. Oceanogr. 57, 1517–1526. doi:10.4319/lo.2012.57.5.1517

Bahr, M., Crump, B. C., Klepac-Ceraj, V., Teske, A., Sogin, M. L., and Hobbie,J. E. (2005). Molecular characterization of sulfate-reducing bacteria in a NewEngland salt marsh. Environ. Microbiol. 7, 1175–1185. doi: 10.1111/j.1462-2920.2005.00796.x

Baptista, A. M., Seaton, C., Wilkin, M., Riseman, S., Needoba, J. A., Maier, D.,et al. (2015). Infrastructure for collaborative estuarine science and societalapplications in the Columbia River, United States. Front. Earth Sci. doi:10.10077/s11707-015-0540-5

Bedoshvili, Y. D., Popkova, T. P., and Likhoshway, Y. V. (2009). Chloroplaststructure of diatoms of different classes. Cell Tissue Biol. 3, 297–310. doi:10.1134/S1990519X09030122

Bersine, K., Brenneis, V. E. F., Draheim, R. C., Rub, A. M. W., Zamon, J. E.,Litton, R. K., et al. (2008). Distribution of the invasive New Zealand mudsnail(Potamopyrgus antipodarum) in the Columbia River Estuary and its firstrecorded occurrence in the diet of juvenile Chinook salmon (Oncorhynchustshawytscha). Biol. Invasions 10, 1381–1388. doi: 10.1007/s10530-007-9213-y

Bianchi, T. S. (2011). The role of terrestrially derived organic carbon in the coastalocean. A changing paradigm and the priming effect. Proc. Natl. Acad. Sci. U.S.A.108, 19473–19481. doi: 10.1073/pnas.1017982108

Blum, L. K., and Mills, A. L. (2012). “Estuarine microbial ecology,” in EstuarineEcology, 2nd Edn, eds J. W. Jr. Day, B. C. Crump, W. M. Kemp, and A.Yanez-Arancibia (Hoboken, NJ: Wiley-Blackwell), 235–263.

Borrel, G., O’Toole, P. W., Harris, H. M. B., Peyret, P., Brugere, J.-F.,and Gribaldo, S. (2013). Phylogenomic data support a seventh orderof methylotrophic methanogens and provide insights into the evolutionof methanogenesis. Genom. Biol. Evol. 5, 1769–1780. doi: 10.1093/gbe/evt128

Bottom, D. L., Simenstad, C. A., Burke, J., Baptista, A. M., Jay, D. A., Jones, K. K.,et al. (2005). Salmon at River’s End, The Role of the Estuary in the Decline andRecovery of Columbia River Salmon. Technical Memorandum NMFS-NWFSC-68. Seattle, WA: U.S. Department of Commerce, NOAA, 1–246.

Bräuer, S. L., Adams, C., Kranzler, K., Murphy, D., Xu, M., Zuber, P., et al. (2011).Culturable Rhodobacter and Shewanella species are abundant in estuarineturbidity maxima of the Columbia River. Environ. Microbiol. 13, 589–603. doi:10.1111/j.1462-2920.2010.02360.x

Breckenridge, J. K., Bollens, S. M., Rollwagen-Bollens, G., and Roegner, G. C.(2014). Plankton assemblage variability in a river-dominated temperate estuaryduring late spring (high-flow) and late summer (low-flow) periods. EstuariesCoasts 38, 93–103. doi: 10.1007/s12237-014-9820-7

Cai, W.-J., Pomeroy, L. R., Moran, M. A., and Wang, Y. (1999). Oxygen andcarbon dioxide mass balance for the estuarine-intertidal marsh complex offive rivers in the southeastern U. S. Limnol. Oceanogr. 44, 639–649. doi:10.4319/lo.1999.44.3.0639

Caspi, R., Altman, T., Dreher, K., Fulcher, C. A., Subhraveti, P., Keseler, I. M.,et al. (2012). The MetaCyc database of metabolic pathways and enzymes andthe BioCyc collection of pathway/genome databases. Nucleic Acids Res. 40,D742–D753. doi: 10.1093/nar/gkr1014

Chawla, A., Jay, D., Baptista, A., Wilkin, M., and Seaton, C. (2008). Seasonalvariability and estuary–shelf interactions in circulation dynamics of a river-dominated estuary. Estuaries Coasts 31, 269–288. doi: 10.1007/s12237-007-9022-7

Chistoserdova, L., Jenkins, C., Kalyuzhnaya, M. G., Marx, C. J., Lapidus, A.,Vorholt, J. A., et al. (2004). The enigmatic Planctomycetes may hold a key to theorigins of methanogenesis and methylotrophy. Mol. Biol. Evol. 21, 1234–1241.doi: 10.1093/molbev/msh113

Chistoserdova, L., Vorholt, J. A., Thauer, R. K., and Lidstrom, M. E.(1998). C1 Transfer enzymes and coenzymes linking methylotrophic bacteriaand methanogenic archaea. Science 281, 99–102. doi: 10.1126/science.281.5373.99

Crump, B. C., Armbrust, E. V., and Baross, J. A. (1999). Phylogenetic analysisof particle-attached and free-living bacterial communities in the Columbiariver, its estuary, and the adjacent coastal ocean. Appl. Environ. Microbiol. 65,3192–3204.

Crump, B. C., Baross, J. A., and Simenstad, C. A. (1998). Dominance of particle-attached bacteria in the Columbia River estuary. USA. Aquat. Microb. Ecol. 14,7–18. doi: 10.3354/ame014007

Crump, B. C., Ducklow, H.W., and Hobbie, J. E. (2012). “Estuarine microbial foodwebs,” in Estuarine Ecology, 2nd Edn, eds J. W. Jr. Day, B. C. Crump, W. M.Kemp, and A. Yanez-Arancibia (Hoboken, NJ: Wiley-Blackwell), 263–284.

Crump, B. C., Hopkinson, C. S., Sogin, M. L., and Hobbie, J. E. (2004). Microbialbiogeography along an estuarine salinity gradient, combined influences ofbacterial growth and residence time. Appl. Environ. Microbiol. 70, 1494–1505.doi: 10.1128/AEM.70.3.1494-1505.2004

Delhez, E. J. M., and Deleersnijder, E. (2002). The concept of age in marinemodelling. II. Concentration distribution function in the English Channel andthe North Sea. J. Marine Syst. 31, 279–297. doi: 10.1111/1755-0998.12371

DeLong, E. F., Franks, D. G., and Alldredge, A. L. (1993). Phylogenetic diversityof aggregate-attached vs. free-living marine bacterial assemblages. Limnol.Oceanogr. 38, 924–934. doi: 10.4319/lo.1993.38.5.0924

Dridi, B. D., Fardeau, M.-L., Ollivier, B., Raoult, D., and Drancourt, M. (2012).Methanomassiliicoccus luminyensis gen. nov., sp. nov., a methanogenicarchaeon isolated from human faeces. Int. J. Syst. Evol. Microbiol. 62, 1902–1907. doi: 10.1099/ijs.0.033712-0

Finn, R. D., Clements, J., and Eddy, S. R. (2011). HMMER web server,interactive sequence similarity searching. Nucleic Acids Res. 39, W29–W37. doi:10.1093/nar/gkr367

Frey, B. E., Lara-Lara, R., and Small, L. F. (1984). Water Column PrimaryProduction in the Columbia River Estuary. Portland, OR: Columbia RiverEstuary Data Development Program.

Gilbert, M., Needoba, J., Koch, C., Barnard, A., and Baptista, A. (2013). Nutrientloading and transformations in the Columbia River estuary determinedby high-resolution in situ sensors. Estuaries Coasts 36, 708–727. doi: 10.1007/s12237-013-9597-0

Grossart, H.-P., Tang, K. W., Kiorboe, T., and Ploug, H. (2007). Comparison ofcell-specific activity between free-living and attached bacteria using isolates andnatural assemblages. FEMS Microbiol. Lett. 266, 194–200. doi: 10.1111/j.1574-6968.2006.00520.x

Guenet, B., Danger, M., Abbadie, L., and Lacroix, G. (2010). Priming effect,bridging the gap between terrestrial and aquatic ecology. Ecology 91, 2850–2861.doi: 10.1890/09-1968.1

Haertel, L. S., Osterberg, C. L., Curl, H., and Park, P. K. (1969). Nutrient andplankton ecology of the Columbia River estuary. Ecology 50, 962–978. doi:10.2307/1936889






Herfort, L., Peterson, T. D., McCue, L. A., Crump, B. C., and Baptista, A. M.(2011a). Myrionecta rubra population genetic diversity and its cryptophytechloroplast specificity in recurrent red tides in the Columbia River estuary.Aquat. Microb. Ecol. 62, 85–97. doi: 10.3354/ame01460

Herfort, L., Peterson, T. D., McCue, L. E., and Zuber, P. (2011b). Protist 18S rRNAgene sequence analysis reveals multiple sources of organic matter contributingto turbidity maxima of the Columbia River estuary. Mar. Ecol. Prog. Ser. 438,19–31. doi: 10.3354/meps09303

Jiao, N., Herndl, G. J., Hansell, D. A., Benner, R., Kattner, G., Wilhelm, S. W., et al.(2010). Microbial production of recalcitrant dissolved organic matter, long-term carbon storage in the global ocean. Nat. Rev. Microbiol. 8, 593–599. doi:10.1038/nrmicro2386

Jickells, T. D. (1998). Nutrient biogeochemistry of the coastal zone. Science 281,217–222. doi: 10.1126/science.281.5374.217

Johnson, M., Zaretskaya, I., Raytselis, Y., Merezhuk, Y., McGinnis, S., and Madden,T. L. (2008). NCBI BLAST, a better web interface. Nucleic Acids Res. 36,W5–W9. doi: 10.1093/nar/gkn201

Kalyuzhnaya, M. G., Hristova, K. R., Lidstrom, M. E., and Chistoserdova, L.(2008). Characterization of a novel methanol dehydrogenase in representativesof Burkholderiales, implications for environmental detection of methylotrophyand evidence for convergent evolution. J. Bacteriol. 190, 3817–3823. doi:10.1128/JB.00180-08

Kärnä, T., Baptista, A. M., Lopez, J. E., Turner, P. J., McNeil, C., andSanford, T. B. (2015). Numerical modeling of circulation in high-energyestuaries: a Columbia River Estuary benchmark. Ocean Model. 88, 54–71. doi:10.1016/j.ocemod.2015.01.001

Kirchman, D. L. (2002). The ecology of Cytophaga–Flavobacteria in aquaticenvironments. FEMS Microbiol. Ecol. 39, 91–100. doi: 10.1111/j.1574-6941.2002.tb00910.x

Klinkhammer, G. P., and McManus, J. (2001). Dissolved manganese in theColumbia River estuary, production in the water column.Geochim. Cosmochim.Acta 65, 2835–2841. doi: 10.1016/S0016-7037(01)00650-0

Kneifel, H. (1979). “Amines in Algae,” inMarine Algae in Pharmaceutical Science,Vol. 1, eds H. A. Hoppe, T. Levring, and Y. Tanaka (Berlin: deGruyterPublishing), 365–402.

Knittel, K., and Boetius, A. (2009). Anaerobic oxidation of methane,progress with an unknown process. Ann. Rev. Microbiol. 63, 311–334.doi: 10.1146/annurev.micro.61.080706.093130

Konstantinidis, K. T., Braff, J., Karl, D. M., and DeLong, E. F. (2009).Comparative metagenomic analysis of a microbial community residingat a depth of 4,000 meters at station ALOHA in the North PacificSubtropical Gyre. Appl. Environ. Microbiol. 75, 5345–5355. doi: 10.1128/AEM.00473-09

Kubo, K., Lloyd, K. G., Biddle, J. F., Amann, R., Teske, A., and Knittel, K.(2012). Archaea of the Miscellaneous Crenarchaeotal Group are abundant,diverse and widespread in marine sediments. ISME J. 6, 1949–1965. doi:10.1038/ismej.2012.37

Kudela, R., Pitcher, G., Probyn, T., Figueiras, F., Moita, T., and Trainer, V. (2005).Harmful algal blooms in coastal upwelling systems. Oceanography 18, 184–197.doi: 10.5670/oceanog.2005.53

Lazar, C. S., John Parkes, R., Cragg, B. A., L’Haridon, S., and Toffin, L. (2012).Methanogenic activity and diversity in the centre of the Amsterdam MudVolcano. Eastern Mediterranean Sea. FEMS Microbiol. Ecol. 81, 243–254. doi:10.1111/j.1574-6941.2012.01375.x

Leloup, J., Fossing, H., Kohls, K., Holmkvist, L., Borowski, C., and Jørgensen,B. B. (2009). Sulfate-reducing bacteria in marine sediment (Aarhus Bay.Denmark), abundance and diversity related to geochemical zonation.Environ. Microbiol. 11, 1278–1291. doi: 10.1111/j.1462-2920.2008.01855.x

Lev, E., Vander Schaaf, D., Anderson, J., Christy, J., Adamus, P., Popper, K.,et al. (2008). The Youngs Bay Watershed Conservation and Restoration Plan.Portland, OR: The Wetlands Conservancy.

Lloyd, K. G., Schreiber, L., Petersen, D. G., Kjeldsen, K. U., Lever, M. A., Steen,A. D., et al. (2013). Predominant archaea in marine sediments degrade detritalproteins. Nature 496, 215–218. doi: 10.1038/nature12033

Macdonald, K. B., and Winfield, T. P. (1984). Tidal Marsh Plant Productionin the Columbia River Estuary. Portland, OR: Columbia River Estuary DataDevelopment Program.

Markowitz, V. M., Chen, I. M. A., Palaniappan, K., Chu, K., Szeto, E., Grechkin, Y.,et al. (2012). IMG, the integrated microbial genomes database and comparativeanalysis system. Nucleic Acids Res. 40, D115–D122. doi: 10.1093/nar/gkr1044

Markowitz, V. M., Szeto, E., Palaniappan, K., Grechkin, Y., Chu, K., Chen, I. M. A.,et al. (2008). The integrated microbial genomes, data content and analysis toolextensions.Nucleic Acids Res. 36, D528–D533. doi: 10.1093/nar/gkm846

Meyerdierks, A., Kube, M., Lombardot, T., Knittel, K., Bauer, M., Glöckner, F. O.,et al. (2005). Insights into the genomes of archaea mediating the anaerobicoxidation of methane. Environ. Microbiol. 7, 1937–1951. doi: 10.1111/j.1462-2920.2005.00844.x

Middelburg, J., Nieuwenhuize, J., Iversen, N., Hogh, N., de Wilde, H., Helder, W.,et al. (2002). Methane distribution in European tidal estuaries. Biogeochemistry59, 95–119. doi: 10.1023/A:1015515130419

Narihiro, T., and Sekiguchi, Y. (2007). Microbial communities in anaerobicdigestion processes for waste and wastewater treatment, a microbiologicalupdate. Curr. Opin. Biotechnol. 18, 273–278. doi: 10.1016/j.copbio.2007.04.003

Narihiro, T., Terada, T., Ohashi, A., Kamagata, Y., Nakamura, K., and Sekiguchi, Y.(2012). Quantitative detection of previously characterized syntrophic bacteriain anaerobic wastewater treatment systems by sequence-specific rRNA cleavagemethod.Water Res. 46, 2167–2175. doi: 10.1016/j.watres.2012.01.034

Odum, E. P. (2000). “Tidal marshes as outwelling/pulsing systems,” inConcepts andControversies in Tidal Marsh Ecology, eds M. P. Weinstein and D. A. Kreeger(Berlin: Kluwer Academic Publishers), 3–7.

Oremland, R. S., and Polcin, S. (1982). Methanogenesis and sulfate reduction,competitive and noncompetitive substrates in estuarine sediments. Appl.Environ. Microbiol. 44, 1270–1276.

O’Sullivan, L. A., Sass, A. M., Webster, G., Fry, J. C., Parkes, R. J., andWeightman, A. J. (2013). Contrasting relationships between biogeochemistryand prokaryotic diversity depth profiles along an estuarine sediment gradient.FEMS Microbiol. Ecol. 85, 143–157. doi: 10.1111/1574-6941.12106

Paul, K., Nonoh, J. O., Mikulski, L., and Brune, A. (2012). Methanoplasmatales.Thermoplasmatales-related archaea in termite guts and other environments, arethe seventh order of methanogens.Appl. Environ. Microbiol. 78, 8245–8253. doi:10.1128/AEM.02193-12

Pell, J., Hintze, A., Canino-Koning, R., Howe, A., Tiedje, J. M., and Brown,C. T. (2012). Scaling metagenome sequence assembly with probabilisticde Bruijn graphs. Proc. Natl. Acad. Sci. U.S.A. 109, 13272–13277. doi:10.1073/pnas.1121464109

Peng, Y., Leung, H. C.M., Yiu, S.M., and Chin, F. Y. L. (2012). IDBA-UD, a de novoassembler for single-cell and metagenomic sequencing data with highly unevendepth. Bioinformatics 28, 1420–1428. doi: 10.1093/bioinformatics/bts174

Plugge, C. M., Scholten, J. C. M., Culley, D. E., Nie, L., Brockman, F. J.,and Zhang, W. (2010). Global transcriptomics analysis of the Desulfovibriovulgaris change from syntrophic growth with Methanosarcina barkeri tosulfidogenic metabolism. Microbiology 156, 2746–2756. doi: 10.1099/mic.0.038539-0

Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., et al. (2013). TheSILVA ribosomal RNA gene database project, improved data processing andweb-based tools. Nucleic Acids Res. 41, D590–D596. doi: 10.1093/nar/gks1219

Raes, J., Korbel, J., Lercher, M., von Mering, C., and Bork, P. (2007). Predictionof effective genome size in metagenomic samples. Genome Biol. 8, R10. doi:10.1186/gb-2007-8-1-r10

Roegner, G. C., Needoba, J. A., and Baptista, A. M. (2011a). Coastal upwellingsupplies oxygen-depleted water to the Columbia River estuary. PLoS ONE6:e18672. doi: 10.1371/journal.pone.0018672

Roegner, G. C., Seaton, C., and Baptista, A. M. (2011b). Climatic and tidal forcingof hydrography and chlorophyll concentrations in the Columbia River estuary.Estuaries Coasts 34, 281–296. doi: 10.1007/s12237-010-9340-z

Sansone, F. J., Holmes,M. E., and Popp, B. N. (1999).Methane stable isotopic ratiosand concentrations as indicators of methane dynamics in estuaries. GlobalBiogeochem. Cycles 13, 463–474. doi: 10.1029/1999GB900012

Satinsky, B. M., Crump, B. C., Smith, C. B., Sharma, S., Zielinski, B. L., Doherty, M.,et al. (2014). Microspatial gene expression patterns in the Amazon RiverPlume. Proc. Natl. Acad. Sci. U.S.A. 111, 11085–11090. doi: 10.1073/pnas.1402782111

Sherwood, C., Creager, J. S., Roy, E. H., Gelfenbaum, G., and Dempsey, T.(1984). Sedimentary Processes and Environments in the Columbia River Estuary.Portland, OR: Columbia River Estuary Data Development Program.






Simenstad, C. A., Jay, D. A., McIntire, C. D., Nehlsen,W., Sherwood, C., and Small,L. F. (1984).The Dynamics of the Columbia River Estuarine Ecosystem. Portland,OR: Columbia River Estuary Data Development Program.

Simenstad, C. A., Small, L. F., McIntire, C. D., Jay, D. A., and Sherwood, C. (1990).Columbia river estuary studies. An introduction to the estuary, a brief history,and prior studies. Prog. Oceanogr. 25, 1–13.

Simon, H. M., Smith, M. W., and Herfort, L. (2014). Metagenomic insights intoparticles and their associated microbiota in a coastal margin ecosystem. Front.Microbiol. 5:466. doi: 10.3389/fmicb.2014.00466

Simon, M., Grossart, H.-P., Schweitzer, B., and Ploug, H. (2002). Microbial ecologyof organic aggregates in aquatic ecosystems. Aquat. Microb. Ecol. 28, 175–211.doi: 10.1007/s00248-012-0147-2

Smith, M. W., Herfort, L., Tyrol, K., Suciu, D., Campbell, V., and Crump, B. C.(2010). Seasonal changes in bacterial and archaeal gene expression patternsacross salinity gradients in the Columbia River coastal margin. PLoS ONE5:e13312. doi: 10.1371/journal.pone.0013312

Smith, M. W., Zeigler Allen, L., Allen, A. E., Herfort, L., and Simon, H. M.(2013). Contrasting genomic properties of free-living and particle-attachedmicrobial assemblages within a coastal ecosystem. Front. Microbiol. 4:120. doi:10.3389/fmicb.2013.00120

Sullivan, B. E., Prahl, F. G., Small, L. F., and Covert, P. A. (2001). Seasonality ofphytoplankton production in the Columbia River. A natural or anthropogenicpattern? Geochim. Cosmochim. Acta 65, 1125–1139. doi: 10.1016/S0016-7037(00)00565-2

Teeling, H., Fuchs, B. M., Becher, D., Klockow, C., Gardebrecht, A., Bennke,C. M., et al. (2012). Substrate-controlled succession of marine bacterioplanktonpopulations induced by a phytoplankton bloom. Science 336, 608–611. doi:10.1126/science.1218344

Turner, A., and Millward, G. E. (2002). Suspended particles, their role inestuarine biogeochemical cycles. Estuarine Coast. Shelf Sci. 55, 857–883. doi:10.1111/gcb.12923

Williams, T. J., Long, E., Evans, F., DeMaere,M. Z., Lauro, F. M., Raftery, M. J., et al.(2012). A metaproteomic assessment of winter and summer bacterioplankton

from Antarctic Peninsula coastal surface waters. ISME J. 6, 1883–1900. doi:10.1038/ismej.2012.28

Xu, M., Schnorr, J., Keibler, B., and Simon, H. M. (2012). Comparative analysis of16S rRNA and amoA genes from archaea selected with organic and inorganicamendments in enrichment culture. Appl. Environ. Microbiol. 78, 2137–2146.doi: 10.1128/AEM.06845-11

Youngblut, N. D., Wirth, J. S., Henriksen, J. R., Smith, M., Simon, H.,Metcalf, W. W., et al. (2015). Genomic and phenotypic differentiation amongMethanosarcina mazei populations fromColumbia River sediment. ISME J. doi:10.1038/ismej.2015.31 [Epub ahead of print].

Zeleke, J., Lu, S.-L., Wang, J.-G., Huang, J.-X., Li, B., and Ogram, A. (2013). Methylcoenzyme M reductase A (mcrA) gene-based investigation of methanogensin the mudflat sediments of Yangtze River estuary, China. Microb. Ecol. 66,257–267. doi: 10.1007/s00248-012-0155-2

Zhang, Y., and Baptista, A. M. (2008). SELFE, a semi-implicit Eulerian-Lagrangianfinite-element model for cross-scale ocean circulation. OceanModel. 21, 71–96.doi: 10.1016/j.ocemod.2007.11.005

Zimmermann-Timm, H. (2002). Characteristics, dynamics and importanceof aggregates in rivers – an invited review. Int. Rev. Hydrobiol. 87,197–240. doi: 10.1002/1522-2632(200205)87:2/3<197::AID-IROH197>3.0.CO;2-7

Conflict of Interest Statement: The authors declare that the research wasconducted in the absence of any commercial or financial relationships that couldbe construed as a potential conflict of interest.

Copyright © 2015 Smith, Davis, Youngblut, Kärnä, Herfort, Whitaker, Metcalf,Tebo, Baptista and Simon. This is an open-access article distributed under the termsof the Creative Commons Attribution License (CC BY). The use, distribution orreproduction in other forums is permitted, provided the original author(s) or licensorare credited and that the original publication in this journal is cited, in accordancewith accepted academic practice. No use, distribution or reproduction is permittedwhich does not comply with these terms.










metagenomic evidence for reciprocal particle exchange ... · smith et al. columbia river estuarine...

Documents